absurd

name	absurd
description	Postgres-native durable workflow system — run workflows as SQL stored procedures, no extra infrastructure.
tags	["durable-execution","postgres","workflows","python","orchestration"]
related_skills	["hermes-agent"]

Absurd — Postgres-Native Durable Workflows

Absurd runs durable workflows entirely within PostgreSQL using stored procedures. No Temporal, no Prefect, no Celery — just Postgres.

Prerequisites

• Python 3.10+
• PostgreSQL 14+ (running and accessible)

Installation

pip install absurd

Quick Start

# Initialize in your Postgres database
absurd init --database postgresql://user:***@localhost:5432/myapp

# Define a workflow
absurd workflow create process_order << 'SQL'
INSERT INTO absurd.workflows (name, steps) VALUES
  ('process_order', ARRAY['validate', 'charge', 'ship']);
SQL

# Execute
absurd run process_order --input '{"order_id": 42}'

# Check status
absurd status <execution-id>

Common Patterns

Retry with backoff

-- Workflows automatically retry failed steps
-- Configure via workflow definition
absurd workflow update process_order --max-retries 3 --backoff exponential

Monitoring

# List all running workflows
absurd list --status running

# View execution history
absurd history --workflow process_order --limit 10

Pitfalls

• Requires pg_cron extension for scheduled workflows
• Stored procedures use SECURITY DEFINER — ensure Postgres user has appropriate permissions
• Long-running workflows may hold connections — configure connection pooling

Verification

# Verify installation
absurd --version

# Verify database connection
absurd status --database postgresql://localhost/myapp